Gas Chromatographic Determination of Formaldehyde by Sucrose Octa-Acetate-Columpak T R. S. Mann and K. W. Hahn Department of Chemical Engineering, Unioersity of Ottawa, Ottawa 2, Canada
FORMALDEHYDE is usually produced commercially by the heterogeneous oxidation of methanol, and as such the separations of formaldehyde from methanol and water are of major importance. Recently, several workers have reported the separation of formaldehyde from methanol by gas chromatography ; the more interesting separations are those of McReynolds ( I ) , Bombaugh and Bull (2), and Smolkova et al. (3). McReynolds (1) reported separations on columns using sucrose octa-acetate supported on Celite, which had been previously washed with acetic acid. He observed that methanol interfered with the elution of formaldehyde. Bombaugh and Bull (2), who tested eight substrates and five solid supports using four different types of gas chromatographs for determining formaldehyde in solution or as a high purity gas, recommended a 500-cm, 6.35-mm copper column with 10 wt % Ethofat 60/25 on Columpak T. Smolkova et al. (3) tried seven substrates on two solid supports, and found that though formaldehyde, water, and methanol could be determined both quantitatively and qualitatively using 30 wt % Reoplex on Celite, 28 wt % PEGA on Rysorb, and 28 PEGA on Celite; the last gave the best results and wt methanol eluted much before formaldehyde. Porapak N may also be used to separate formaldehyde from methanol and water. However, it usually produces a greater pressure drop across the column than Columpak T , and therefore is not suitable in cases where long columns are required. In studying the kinetics of the oxidation of methanol t o formaldehyde, the latter is the more important product. A sharp symmetrical peak with a short retention time is desirable, and no advantage is to be gained by having the formaldehyde elute after methanol and water. This work describes the use of 15 wt % sucrose octa-acetate on Columpak T in a 500-cm, 6.35-mm copper tube and compares the results with 10 wt % Ethofat 60/25 on Columpak T , with 28 wt % PEGA on Celite, and with Porapak N. Both gas density and thermal conductivity detectors were used. Although the results with both were nearly the same, the latter was more sensitive. EXPERIMENTAL Apparatus. The chromatogram presented here (Figure 1) was obtained on gas chromatographic equipment constructed in this laboratory. The column was immersed in a liquid bath (Fisher bath oil No. 2 ) . Its temperature was kept constant with a Yellow Springs thermistor temperature controller, Model 71, having a controlled temperature deviation within 10.01 "C. A Gow-Mac, Model T R IIIA, 4 W 2, temperature-regulated thermal conductivity cell, and a Gow-Mac thermal gas density detector were used. The temperature of the injection block was maintained a t 200" C. Other parts of the assembly were wrapped in heating tapes covered with asbestos sheets. The heating tapes were controlled by a Model 63 (1) W. 0. McReynolds, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, 1961. (2) K. J. Bombaugh and W. C. Bull, ANAL.CHEM., 34, 1237 (1962). ( 3 ) E. Smolkova, V. Kolouskova, and L. Feltl, 2. Anal. Clzem., 202,262 (1964).
13 14
ANALYTICAL CHEMISTRY
Figure 1. Chromatographic separation on 15 wt% sucrose octa-acetate-Columpak T at 93" C Column: Length 500 cm, inlet pressure 872 (20 psig) mm Hg/(at room temp.), outlet pressure 769 mm Hg. Sample size: 1 p1 of solution containing 37.2 % formaldehyde, 13.1 % methanol,and 49.7 water. Bridge current: 250 mA. Sensitivity setting: 32. Carrier gas flow rate: 54 ml/minute. Room temperature: 24.5"C. Detector: thermal conductivity-filaments. Carrier gas: Helium 1. Formaldehyde 2. Methanol 3. Water
Yellow Springs temperature controller, with a temperature deviation of 10.05" C. Interconnections used were made from 0.125-inch stainless-steel tubing, Liquid samples were injected with a Hamilton microliter syringe. Reagents. The substrates tested were sucrose octaacetate, C28H38019, M. W. 678.59, Eastman Kodak grade (Lot No. P4024, Canlab Catalog No. V-134); and Ethofat 60/25 (Lot No. 8174, Armour Industrial Chemical Co.), a polyethylene monostearate, containing a n average of 15 ethylene oxide units, M. W. 938. The solid supports used were Columpak T (Lot No. 724875, Fisher Scientific CO. Catalog No. C-587), and Porapak N, 80-100 mesh (Batch No, 500, Waters Associates Inc.). Methylene chloride reagent grade was used as a solvent for all the substrates. Procedure. Procedures similar to those described by Bombaugh and Bull (2) were followed. The Porapak N column was dried overnight at 200" C before use, and the
sucrose octa-acetate column at 195" C for 1 hour. The flow rate of carrier gas was checked by both the soap bubble method and "Fisher Wet Test" gas meter. Separations were also checked on a n Aerograph Model 90-P3 gas chromatograph with filament detector, a Perkin-Elmer Model 154C with thermistor and the departmental Gow-Mac 4W2 gas chromatographs. Similar curves were obtained on all three instruments. RESULTS AND DISCUSSION
The theoretical plate (n) requirement of each particular separation has been calculated according t o the method of Kaiser ( 4 ) and defined as: n = 5.54
(5J
(1)
where t d r = retention time, (uncorrected), measured in mm or seconds, and blI2 = peak width a t half height, measured in m m or seconds. The minimum detectable quantity of sample (MDQ) is defined as the quantity required to produce 0.1-mV signal on a I-mV recorder, when the sensitivity setting of the gas chromatograph is 1, MDQ
=
x
(mg sample) (wt sample) (0.1 mV) (2) (sensitivity of gas chromatograph) (height of peak in mV)
most efficient in the elution and separation of formaldehyde from methanol and water as compared to other substrates tested. Sucrose octa-acetate, 15 wt %, on Columpak T gave the lowest tdr/bl,2values (4.9 a t 93" c and 4.1 a t 97" C), the smallest number of theoretical plates (135 a t 93" C and 92 a t 97 " C), and the greatest sensitivity (MDQ-formaldehyde = 0.0020 mg at 93 " C and 0.0009 a t 97" C) of all the columns tested. The retention time of formaldehyde was 5.2 minutes a t 93°C (Figure 1) and 4 minutes a t 97" C. The substrate for this column could be easily loaded on its support up to 20 by weight, and still pack without difficulty. It could also be used a t operating temperatures less than 93" C , and was stable over a wide temperature range. As compared t o sucrose octa-acetate on Columpak T, the tdr/blr2 values for formaldehyde with 10 wt Ethofat 60/25 PEGA on Celite on Columpak T at 115" C , with 20 wt a t 110" C (3), and with Porapak N a t 110" C were 5.4, 5.7, and 14.5, respectively. Similarly the smallest number of theoretical plates were, respectively, in that order 160, 179, and 1162. MDQ-formaldehyde, in mg for Ethofat 60/25, and Porapak N were 0.0525 and 0.0441, respectively. ACKNOWLEDGMENT
Sucrose octa-acetate heated a t 195" C for 1 hour proved
The authors express their appreciation to K . J. Bombaugh, Mine Safety Appliances, Pittsburgh, Pa., for helpful suggestions.
(4) R. Kaiser, "Gas Phase Chromatography," Vol. 1, Butterworth and Co., New York, 1963, p. 20.
RECEIVED for review May 12, 1967. Accepted June 23, 1967. Work supported by National Research Council of Canada, Ottawa, Grant A-1125.
Separation and Determination of Trinitrotoluene Isomers by Gas Chromatography D. G. Gehring and J. E. Shirk Eastern Laboratory, Explosices Department, E. I. du Pont de Nemours & Co., Gibbstown, N . J
ANALYTICAL METHODS were required to monitor the individual unsymmetrical T N T isomers in symmetrical 2,4,6-TNT. An early method developed by Halfter ( I ) was based upon the selective reaction between sulfite ion and one of the nitro groups of an unsymmetrical T N T molecule. This method determines the total amount of unsymmetrical T N T but fails to distinguish between the individual unsymmetrical isomers. Subsequently, methods utilizing infrared spectrometry ( 2 , 3), thin layer chromatography ( 4 ) , and paper chromatography ( 5 ) were reported. However, these methods generally lacked sensitivity and/or quantitative accuracy at low concentrations. In due time, gas chromatographic ( G C ) methods appeared which separated the individual dinitrotoluene (DNT) isomers (6, 7), and recently a G C method was reported which de(1) G . Halfter, 2. Anal. Chem., 128, 23 (1947). (2) F. Pristera, Appl. Spectry., 7, No. 3, 115 (1953). (3) F. Pristera, M . Halik, A. Castelli, and W. Fredericks, ANAL. CHEM., 32,495 (1960). (4) S. K. Yasuda, J . Cliromurog., 13,( I ) , 78 (1964). ( 5 ) V. Ettel, S. Pospilsil, and 2. Deyl, Cliem. Lisry, 52, 623 (1958). (6) E. Camera, D. Pravisani, and V. Ohman, Explosicsroffe, Nr 9, 237 (1965). (7) J . S. Parsons, S. M. Tsang, M. P. DiGiairno, R. Feinland, and R . A. L. Paylor, ANAL.CHEM., 33, 1858 (1961).
termines 2,4,6-TNT in castable explosives containing mixtures of T N T and other explosive additives (8). We have succeeded in separating the individual T N T isomers (except for the 2,3,6- and 2,4,6-TNT pair) by gas chromatography. Now it is possible to determine quantitatively the individual components of a n isomeric mixture and to monitor 2,4,6-TNT for low concentrations of unsymmetrical isomers. Also, D N T isomers were separated and may be individually determined together with the T N T components. EXPERIMENTAL
Apparatus. All work was performed using an F & M Scientific Corp., Division of Hewlett-Packard, Model 810 dual-column gas chromatograph equipped with a thermal conductivity filament detector and a 1-mV recorder. Two 9-fOOt X 0.25-inch stainless steel columns (one serving as reference) were loosely packed with 10 DC-LSX-3-0295 silicone copolymer (Applied Science Laboratories Inc., State College, Pa.) on 80-90 mesh Anakrom-ABS support (Analabs Inc., Hamden, Conn.). Both columns were conditioned in the chromatograph for 2 hours a t 250" C oven ~~
~
(8) M. L. Rowe, J . Gus Chromatog.,4, 420 (1966). VOL. 39, NO. 1 1, SEPTEMBER 1967
1315
